scripts/build_data.sh

#!/bin/bash
# Build K-step GUI Transition data from GUI trajectories

set -e

# Default values
INPUT_DIR=""
OUTPUT_DIR="./data/gui_transition"
K_VALUES="1 2 3 4"
SAMPLES_PER_K=2000
SEED=42

# Parse arguments
while [[ $# -gt 0 ]]; do
    case $1 in
        --input_dir)
            INPUT_DIR="$2"
            shift 2
            ;;
        --output_dir)
            OUTPUT_DIR="$2"
            shift 2
            ;;
        --k_values)
            K_VALUES="$2"
            shift 2
            ;;
        --samples_per_k)
            SAMPLES_PER_K="$2"
            shift 2
            ;;
        --seed)
            SEED="$2"
            shift 2
            ;;
        *)
            echo "Unknown option: $1"
            exit 1
            ;;
    esac
done

if [ -z "$INPUT_DIR" ]; then
    echo "Error: --input_dir is required"
    echo "Usage: bash scripts/build_data.sh --input_dir /path/to/trajectories --output_dir ./data/gui_transition"
    exit 1
fi

echo "=== Building K-step GUI Transition Data ==="
echo "Input: $INPUT_DIR"
echo "Output: $OUTPUT_DIR"
echo "K values: $K_VALUES"
echo "Samples per K: $SAMPLES_PER_K"
echo "Seed: $SEED"
echo ""

python src/data_construction/build_kstep_data.py \
    --input_dir "$INPUT_DIR" \
    --output_dir "$OUTPUT_DIR" \
    --k_values $K_VALUES \
    --samples_per_k "$SAMPLES_PER_K" \
    --seed "$SEED"

echo ""
echo "Data construction complete! Output: $OUTPUT_DIR"